Solution for foreign object removal based on spectral imaging is discussed. Both design of classifiers and their real-time deployment to a custom C/C++ application processing live spectral data stream is demonstrated using publicly available release of perClass Mira. The processing speed of 1.45ms/frame (691fps) is achieved on a Linux system with a GPU.
Separating different types of meat (pork, beef and chicken) using Hyspex spectral sensor
Classification of different materials in a live stream of spectral data using Photonfocus camera.
Classifying different complex land cover types such as grass, trees, water, shadows in spectral images acquired from a drone (Hyspex)
We gave a live demo of white & transparent plastic detection in (white) food product at Agri&Food Tech event in Den Bosch, The Netherlands. (Specim FX17)
Dark textiles can be sorted based on spectral imaging (Imec SWIR SNAPSCAN)
Separating diverse types of nuts and grain using Specim FX17 camera
Detecting red blood cells (Imec VIS-NIR SNAPSCAN)
Detecting virus infection in potato plants using spectral imaging.
Separating different types of plastics using Specim FX17 line-scan camera.
We make software tools for interpretation of spectral images and machine learning solutions